fs/ntfs3: hold ni_lock across readdir metadata walk

[BUG]
KASAN reports a slab-use-after-free during getdents(2):

BUG: KASAN: slab-use-after-free in ntfs_read_mft fs/ntfs3/inode.c:79 [inline]
BUG: KASAN: slab-use-after-free in ntfs_iget5+0x59b/0x3450 fs/ntfs3/inode.c:541
Read of size 2 at addr ffff88800b7a5a4e by task syz.0.1061/2354

Call Trace:
 __dump_stack lib/dump_stack.c:94 [inline]
 dump_stack_lvl+0xbe/0x130 lib/dump_stack.c:120
 print_address_description mm/kasan/report.c:378 [inline]
 print_report+0xd1/0x650 mm/kasan/report.c:482
 kasan_report+0xfb/0x140 mm/kasan/report.c:595
 __asan_report_load2_noabort+0x14/0x30 mm/kasan/report_generic.c:379
 ntfs_read_mft fs/ntfs3/inode.c:79 [inline]
 ntfs_iget5+0x59b/0x3450 fs/ntfs3/inode.c:541
 ntfs_dir_emit fs/ntfs3/dir.c:337 [inline]
 ntfs_read_hdr+0x714/0x930 fs/ntfs3/dir.c:385
 ntfs_readdir+0xaad/0x1010 fs/ntfs3/dir.c:458
 iterate_dir+0x276/0x9e0 fs/readdir.c:108
 __do_sys_getdents fs/readdir.c:326 [inline]
 __se_sys_getdents fs/readdir.c:312 [inline]
 __x64_sys_getdents+0x143/0x290 fs/readdir.c:312
 ...

Allocated by task 2160:
 kasan_save_stack+0x39/0x70 mm/kasan/common.c:56
 kasan_save_track+0x14/0x40 mm/kasan/common.c:77
 kasan_save_alloc_info+0x37/0x60 mm/kasan/generic.c:573
 poison_kmalloc_redzone mm/kasan/common.c:400 [inline]
 __kasan_kmalloc+0xc3/0xd0 mm/kasan/common.c:417
 kasan_kmalloc include/linux/kasan.h:262 [inline]
 __do_kmalloc_node mm/slub.c:5650 [inline]
 __kmalloc_noprof+0x2bd/0x900 mm/slub.c:5662
 kmalloc_noprof include/linux/slab.h:961 [inline]
 mi_init+0x9d/0x110 fs/ntfs3/record.c:105
 mi_format_new+0x6b/0x500 fs/ntfs3/record.c:422
 ni_add_subrecord+0x129/0x540 fs/ntfs3/frecord.c:321
 ntfs_look_free_mft+0x238/0xd90 fs/ntfs3/fsntfs.c:715
 ni_create_attr_list+0x8e6/0x1690 fs/ntfs3/frecord.c:826
 ni_ins_attr_ext+0x5ec/0x9d0 fs/ntfs3/frecord.c:924
 ni_insert_attr+0x2bf/0x830 fs/ntfs3/frecord.c:1091
 ni_insert_resident+0xec/0x3d0 fs/ntfs3/frecord.c:1475
 ni_add_name+0x4b2/0x8a0 fs/ntfs3/frecord.c:2987
 ni_rename+0xa6/0x160 fs/ntfs3/frecord.c:3026
 ntfs_rename+0xa19/0xe00 fs/ntfs3/namei.c:332
 vfs_rename+0xd42/0x1d50 fs/namei.c:5216
 do_renameat2+0x715/0xb60 fs/namei.c:5364
 __do_sys_rename fs/namei.c:5411 [inline]
 __se_sys_rename fs/namei.c:5409 [inline]
 __x64_sys_rename+0x83/0xb0 fs/namei.c:5409
 x64_sys_call+0x8c4/0x26a0 arch/x86/include/generated/asm/syscalls_64.h:83
 do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
 do_syscall_64+0x93/0xf80 arch/x86/entry/syscall_64.c:94
 entry_SYSCALL_64_after_hwframe+0x76/0x7e

Freed by task 85:
 kasan_save_stack+0x39/0x70 mm/kasan/common.c:56
 kasan_save_track+0x14/0x40 mm/kasan/common.c:77
 __kasan_save_free_info+0x3b/0x60 mm/kasan/generic.c:587
 kasan_save_free_info mm/kasan/kasan.h:406 [inline]
 poison_slab_object mm/kasan/common.c:252 [inline]
 __kasan_slab_free+0x6f/0xa0 mm/kasan/common.c:284
 kasan_slab_free include/linux/kasan.h:234 [inline]
 slab_free_hook mm/slub.c:2543 [inline]
 slab_free mm/slub.c:6642 [inline]
 kfree+0x2bf/0x6b0 mm/slub.c:6849
 mi_clear fs/ntfs3/ntfs_fs.h:1107 [inline]
 mi_put+0x10e/0x1a0 fs/ntfs3/record.c:97
 ni_write_inode+0x479/0x2a00 fs/ntfs3/frecord.c:3320
 ntfs3_write_inode+0x51/0x70 fs/ntfs3/inode.c:1042
 write_inode fs/fs-writeback.c:1564 [inline]
 __writeback_single_inode+0x8c9/0xc30 fs/fs-writeback.c:1784
 writeback_sb_inodes+0x5e6/0xf60 fs/fs-writeback.c:2015
 __writeback_inodes_wb+0x10c/0x2d0 fs/fs-writeback.c:2086
 wb_writeback+0x63f/0x900 fs/fs-writeback.c:2197
 wb_check_old_data_flush fs/fs-writeback.c:2301 [inline]
 wb_do_writeback fs/fs-writeback.c:2354 [inline]
 wb_workfn+0x8cc/0xd60 fs/fs-writeback.c:2382
 process_one_work+0x8e0/0x1980 kernel/workqueue.c:3263
 process_scheduled_works kernel/workqueue.c:3346 [inline]
 worker_thread+0x683/0xf80 kernel/workqueue.c:3427
 kthread+0x3f0/0x850 kernel/kthread.c:463
 ret_from_fork+0x50f/0x610 arch/x86/kernel/process.c:158
 ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:245

The faulting address sits 590 bytes inside a freed kmalloc-1k object
allocated by ni_add_subrecord() and freed from ni_write_inode()
writeback.

[CAUSE]
ntfs_readdir() loads all subrecords once, but then drops ni_lock()
before it starts walking the directory metadata through ntfs_read_hdr().
That leaves the current NTFS_DE pointer backed by parent-directory
subrecord memory that concurrent writeback is still allowed to compact
and free.

The later ntfs_dir_emit() -> ntfs_iget5() call exposes the stale e->ref,
but the lifetime bug starts earlier: readdir is still consuming
parent-directory metadata after releasing the lock that protects it.

[FIX]
Keep ni_lock() held from the point where ntfs_readdir() starts
consuming the directory metadata until the walk over root/index entries
is finished.

This closes the parent-directory lifetime hole directly and keeps the
existing readdir d_type behaviour unchanged.

Signed-off-by: ZhengYuan Huang <gality369@gmail.com>
Signed-off-by: Konstantin Komarov <almaz.alexandrovich@paragon-software.com>

Author: Gality369

fs/ntfs3/dir.c

@@ -489,10 +489,17 @@ static int ntfs_readdir(struct file *file, struct dir_context *ctx)
 			goto out;
 	}
 
+	/*
+	 * Keep directory metadata stable for the whole walk. Loading subrecords
+	 * once is not enough if concurrent writeback can still compact ATTR_LIST
+	 * entries and free the record that ntfs_read_hdr() is currently walking.
+	 */
+	ni_lock(ni);
+
 	root = indx_get_root(&ni->dir, ni, NULL, NULL);
 	if (!root) {
 		err = -EINVAL;
-		goto out;
+		goto out_unlock;
 	}
 
 	if (pos >= sbi->record_size) {
@@ -503,7 +510,7 @@ static int ntfs_readdir(struct file *file, struct dir_context *ctx)
 		 */
 		err = ntfs_read_hdr(sbi, ni, &root->ihdr, 0, pos, name, ctx);
 		if (err)
-			goto out;
+			goto out_unlock;
 		bit = 0;
 	}
 
@@ -514,7 +521,7 @@ static int ntfs_readdir(struct file *file, struct dir_context *ctx)
 		/* Get the next used index. */
 		err = indx_used_bit(&ni->dir, ni, &bit);
 		if (err)
-			goto out;
+			goto out_unlock;
 
 		if (bit == MINUS_ONE_T) {
 			/* no more used indexes. end of dir. */
@@ -524,13 +531,13 @@ static int ntfs_readdir(struct file *file, struct dir_context *ctx)
 		if (bit >= max_bit) {
 			/* Corrupted directory. */
 			err = -EINVAL;
-			goto out;
+			goto out_unlock;
 		}
 
 		err = indx_read_ra(&ni->dir, ni, bit << ni->dir.idx2vbn_bits,
 				   &node, &file->f_ra);
 		if (err)
-			goto out;
+			goto out_unlock;
 
 		/*
 		 * Add each name from index in 'ctx'.
@@ -539,9 +546,12 @@ static int ntfs_readdir(struct file *file, struct dir_context *ctx)
 				    ((u64)bit << index_bits) + sbi->record_size,
 				    pos, name, ctx);
 		if (err)
-			goto out;
+			goto out_unlock;
 	}
 
+out_unlock:
+	ni_unlock(ni);
+
 out:
 	kfree(name);
 	put_indx_node(node);